: Despite the availability of precise consensus DMA-binding sequences for hundreds of proteins, and the complete genome sequence of dozens of organisms, it is not possible to predict where a given DNA-binding protein will associate with a genome in vivo. How DNA-binding proteins recognize and bind to a subset of genomic DNA sequences, while at the same time not binding to thousands of computationally indistinguishable sequences, remains a major unsolved problem in biology. Using Saccharomyces cerevisiae as a model system, we have developed a unique set of genetic, biochemical, and genomic tools to attack this problem. . Aim one: Our previous work has shown that in vivo, transcription factors bind to DNA upstream of genes in preference to coding regions, even though both regions contain strong consensus binding sites. Cooperative protein-protein-DNA interactions and differential chromatin accessibility are hypothesized to mediate context-dependent binding. To quantitate the degree of specificity dependent on in vivo factors, and how much is inherent to the protein and DNA, the genome-wide specificities of Raplp and LeuSp will be determined in vitro using purified proteins and naked yeast genomic DNA, and compared to specificity in vivo. Changes in the distribution of Raplp in response to changes in environmental conditions will also be determined. Aim two: Our previous work has shown that nucleosome occupancy throughout the genome is heterogeneous in living yeast cells. We propose experiments to determine the molecular basis for differential nucleosome occupancy, and how it is established, regulated, and maintained in yeast. Aim three: Our data and data from other groups suggest an intimate relationship between target selection by DMA binding proteins (addressed in Aim 1), global chromatin organization (addressed in Aim 2), and transcriptional activity. The third aim specifically tests relationships between these three processes. We will assay Rap1p target selection in strains in which (i) the context of Raplp binding sites has been changed (ii) transcription at specific loci has been disabled, and (iii) the RNA Pol II CTD is mutated. We will map transcription-coupled chromatin modifications and perform high-throughput site-directed mutagenesis to link histone structure and modification to biological outcomes. Human Health: Transcription factors, when missexpressed or mutated, are a prevalent cause of human disease. Better prediction of their in vivo targets may lead to therapies that inhibit binding to inappropriate targets. FAIRE, a new chromatin assay we have developed, has potential as a prognostic or diagnostic tool for diseases (including cancer) that affect, or arise from defects in, chromatin or transcription.